The denomination “mobile telecommunications system” refers generally to any telecommunications system which enables a wireless communication connection between a mobile station (MS) and the fixed parts of the system when the user of the mobile station is moving within the service area of the system. A typical mobile communications system is a Public Land Mobile Network (PLMN). The majority of mobile telecommunications systems in use at the time of the filing of this patent application belong to the second generation of such systems, a well-known example being the GSM system (Global System for Mobile telecommunications). However, the invention also applies to the next or third generation of mobile telecommunications systems, such as a system known as the UMTS (Universal Mobile Telecommunications System) which undergoes standardisation.
Internet real time services have gained popularity during the past few years. IP (Internet Protocol) telephony and different streaming applications are already common in the Internet. Also the demand for wireless access to these real time services is expected to be still growing. Packet switched wireless networks, such as GPRS (General Packet Radio Service), are designed to provide data services, e.g. Internet services, cost effectively. In GPRS the channels are not dedicated for one user continuously but are shared between multiple users. This facilitates efficient data multiplexing. However, GPRS is not originally designed for transferring delay sensitive real time data, e.g. IP telephony sessions. For this reason, GPRS contains various technical solutions that do not meet the requirements set by real time traffic. In the following text, a denomination “delay sensitive data” is used for data flows that should be transferred on real time basis and that may have passive periods during which the data flow is suspended.
In order to better understand the problems of the prior art solutions and the idea of the present invention, the structure of a new generation digital cellular radio system is first shortly described, and GPRS is then described in more detail.
FIG. 1a shows a version of a future cellular radio system which is not entirely new compared to the known GSM system but which includes both known elements and completely new elements. The terminals are connected to the radio access network RAN which includes the base stations and the base station controllers. The core network of a cellular radio system comprises mobile services switching centres (MSC), other network elements (in GSM, e.g. SGSN and GGSN, i.e. Serving GPRS Support Node and Gateway GPRS Support node, where GPRS stands for General Packet Radio Service) and related transmission systems. According, e.g. to the GSM+ specifications developed from GSM, the core network can also provide new services.
In FIG. 1a, the core network of a cellular radio system 10 comprises a GSM+ core network 11 which has three parallel radio access networks linked to it. Of those, networks 12 and 13 are UMTS radio access networks and network 14 is a GSM+ radio access network. The upper UMTS radio access network 12 is, e.g. a commercial radio access network, owned by a telecommunications operator offering mobile services, which equally serves all subscribers of said telecommunications operator. The lower UMTS radio access network 13 is, e.g. private and owned e.g. by a company in whose premises said radio access network operates. Typically the cells of the private radio access network 13 are nano- and/or picocells in which only terminals of the employees of said company can operate. All three radio access networks may have cells of different sizes offering different types of services. Additionally, cells of all three radio access networks 12, 13 and 14 may overlap either entirely or in part. The bit rate used at a given moment of time depends, among other things, on the radio path conditions, characteristics of the services used, regional overall capacity of the cellular system and the capacity needs of other users. The new types of radio access networks mentioned above are called generic radio access networks (GRAN). Such a network can co-operate with different types of fixed core networks CN and especially with the GPRS network of the GSM system. The generic radio access network (GRAN) can be defined as a set of base stations (BS) and radio network controllers (RNC) that are capable of communicating with each other using signaling messages.
FIG. 1b shows an architecture of a general packet radio service (GPRS). The GPRS is a new service that is currently based on the GSM system but it is supposed to be generic in the future. GPRS is one of the objects of the standardisation work of the GSM phase 2+ and UMTS at the ETSI (European Telecommunications Standards Institute). The GPRS operational environment comprises one or more subnetwork service areas, which are interconnected by a GPRS backbone network. A subnetwork comprises a number of packet data service nodes (SN), which in this application will be referred to as serving GPRS support nodes (SGSN) 153, each of which is connected to the mobile telecommunications system (typically to a base station through an interworking unit) in such a way that it can provide a packet service for mobile data terminals 151 via several base stations 152, i.e. cells. The intermediate mobile communication network provides packet-switched data transmission between a support node and mobile data terminals 151. Different subnetworks are in turn connected to an external data network, e.g. to a Public Data Network (PDN) 155, via GPRS gateway support nodes GGSN 154. The GPRS service thus allows the provision of packet data transmission between mobile data terminals and external data networks when the appropriate parts of a mobile telecommunications system function as an access network.
In order to access the GPRS services, a mobile station shall first make its presence known to the network by performing a GPRS attachment. This operation establishes a logical link between the mobile station and the SGSN, and makes the mobile station available for SMS (Short Message Services) 158, 159, over GPRS, paging via SGSN, and notification of incoming GPRS data. More particularly, when the mobile station attaches to the GPRS network, i.e. in a GPRS attachment procedure, the SGSN creates a mobility management context (MM context). Also the authentication of the user is carried out by the SGSN in the GPRS attachment procedure. In order to send and receive GPRS data, the MS shall activate the packet data address wanted to be used, by requesting a PDP activation procedure (Packet Data Protocol). This operation makes the mobile station known in the corresponding GGSN, and interworking with external data networks can commence. More particularly, a PDP context is created in the mobile station and the GGSN and the SGSN. The packet data protocol context defines different data transmission parameters, such as the PDP type (e.g. X.25 or IP), the PDP address (e.g. X.121 address), the quality of service (QoS) and the NSAPI (Network Service Access Point Identifier). The MS activates the PDP context with a specific message, Activate PDP Context Request, in which it gives information on the TLLI, the PDP type, the PDP address, the required QoS and the NSAPI, and optionally the access point name (APN).
FIG. 1 also shows the following GSM functional blocks: Mobile Switching Center (MSC)/Visitor Location Register (VLR) 160, Home Location Register (HLR) 157 and Equipment Identity Register (EIR) 161. The GPRS system is usually also connected to other Public Land Mobile Networks (PLMN) 156.
Functions applying digital data transmission protocols are usually described as a stack according to the OSI (Open Systems Interface) model, where the tasks of the various layers of the stack, as well as data transmission between the layers, are exactly defined. In the GSM system phase 2+, which in this patent application is observed as an example of a digital wireless data transmission system, there are five operational layers defined.
Relations between the protocol layers are illustrated in FIG. 2. The lowest protocol layer between the mobile station MS and the base station subsystem is the layer 1 (L1) 200, 201, which corresponds to a physical radio connection. Above it, there is located an entity corresponding to the layers 2 and 3 of a regular OSI model, wherein the lowest layer is a radio link control/media access control (RLC/MAC) layer 202, 203; on top of it a PDCP (or LLC) layer 204, 205; and topmost a radio resource control (RRC) layer 206, 207. Between the base station subsystem UTRA BSS of the generic radio access network and an interworking unit/core network IWU/CN located in the core network, there is assumed to be applied a so-called Iu interface, where the layers corresponding to the above described layers from L1 to PDCP are the layers L1 and L2 of the OSI model (blocks 208 and 209 in the drawing), and the layer corresponding to the above described RRC layer is the layer L3 of the OSI model (blocks 210 and 211 in the drawing).
The mobile station MS must include a higher-level control protocol 212 and a protocol 213 for serving higher-level applications, of which the former communicates with the RRC layer 206 in order to realise control functions connected to data transmission connections, and the latter communicates directly with the PDCP layer 204 in order to transmit such data that directly serves the user (for instance digitally encoded speech). In a mobile station of the GSM system, the blocks 212 and 213 are included in the above mentioned MM layer.
In GPRS, a Temporary Block Flow (TBF) is created for transferring data packets on a packet data channel. The TBF is a physical connection used by the two Radio Resource (RR) peer entities to support the unidirectional transfer of higher layer (e.g. LLC or PDCP) Packet Data Units (PDU) on packet data physical channels. The TBF is normally always released when there is no data to be transmitted. This is a problem in voice services because there are silent periods in between active periods. During these silent or “passive” periods no data is transferred and the TBF is thus released. The TBF setup procedure is likely to be too long in order to be set up fast enough when the active period continues.
An example of the resource allocation in the GPRS of the current GSM Phase 2+ specification is next described in more detail.
In the GSM Phase 2+ the uplink resource allocation is currently specified as follows. The Mobile Station (MS) requests uplink radio resources by sending a PACKET CHANNEL REQUEST message to the network. Various access type values are specified for the request message. For data transfer ‘one phase access’, ‘two phase access’ and ‘short access’ access type values are defined. Using ‘short access’ access type value, the MS may request the radio resources to transfer only few RLC data blocks, and therefore it is not applicable for transferring continuous data flows.
When a network receives a PACKET CHANNEL REQUEST message indicating one phase access, it may allocate radio resources on one or several Packet Data Channels (PDCH). The allocation is based on information included in the request message. The following table shows an example for an 11 bit message content of a PACKET CHANNEL REQUEST message:
bitsPacket Channel1110987654321     Access0mmmmmpprrrOne Phase AccessRequest100nnnpprrrShort AccessRequest110000pprrrTwo Phase AccessRequest110001rrrrrPage Response110010rrrrrCell Update110011rrrrrMobility Manage-ment procedure110100rrrrrMeasurementReportAll othersReserved
An 11 bit PACKET CHANNEL REQUEST message indicating one phase access has a field of 5 bits describing the multislot class of the mobile station, a field of two bits indicating requested priority and a field of three bits describing random reference (random mobile station identification information).
The following table shows an example for an 8 bit message content of a PACKET CHANNEL REQUEST message:
bits87654321Packet Channel Access1mmmmmrrOne Phase Access Request00nnnrrrShort Access Request01000rrrTwo Phase Access Request01001rrrPage Response01010rrrCell Update01011rrrMobility Management procedure01100rrrMeasurement ReportAll othersReserved
An 8 bit Packet Channel Request message indicating one phase access has a field of 5 bits describing the multislot class of the mobile station and a field of two bits describing random reference. The information about the allocated radio resources is sent to the Mobile Station with a PACKET UPLINK ASSIGNMENT message.
When a network receives a PACKET CHANNEL REQUEST message indicating two phase access, it may allocate limited radio resources on one packet data channel. The allocated radio resources are transmitted to the mobile station with a PACKET UPLINK ASSIGNMENT message. After this the mobile station transmits a PACKET RESOURCE REQUEST message to the network by using the allocated radio resources. The message defines more accurately the required radio resources, e.g. requested bandwidth and priority, and the radio capability of the mobile station. Based on the information received in the PACKET RESOURCE REQUEST message, the network may assign one or several packet data channels to the TBF and informs the assigned radio resources to the mobile station with a PACKET UPLINK ASSIGNMENT message.
Above, the request of resources was made using the GPRS control channel as an example. There are also other ways of requesting resources in case the cell does not include a GPRS control channel (even if it supports GPRS). In this case the resource request can be made using a GSM control channel.
In the prior art uplink radio resource allocation the following problems may arise:
If the priority field included into the PACKET CHANNEL REQUEST and the PACKET RESOURCE REQUEST messages does not unambiguously define delay sensitive real time traffic, the network might not be able to provide the needed radio resources for the MS. Thus, e.g. transferring speech using the GPRS might not reach a sufficient quality.
The default RLC mode is an acknowledged mode in one phase access. Since real time traffic would be transferred using unacknowledged RLC mode, two phase access should be used. Using two phase access, additional radio resource request information may be given to the network. However, two phase access causes additional delay to channel assignment procedure, because the mobile station has to send two request messages to the network instead of one. In spite of the additional radio resource request information it is not guaranteed that the network is able to provide the needed radio resources for the mobile station transferring delay sensitive real time traffic.
When allocating radio resources for uplink transfer, downlink radio resources cannot be allocated simultaneously, because the downlink temporary block flow cannot be created without downlink packets. Thus it is possible that, when the mobile station then would receive a downlink packet, the network is unable to assign radio resources for the transfer of the packet.
Downlink radio resource allocation is currently specified as follows: When the network receives data for a mobile station which has no assigned radio resources and whose cell location is known, the network assigns radio resources on one or several packet data channels by transmitting a PACKET DOWNLINK ASSIGNMENT message to the mobile station. When the mobile station receives the assignment message, it starts listening allocated packet data channels for Radio Link Control (RLC) data blocks.
In downlink radio resource allocation, the following problems may arise:
If information attached to data (coming from the SGSN) does not unambiguously define delay sensitive real time traffic, the network may not be able to provide the needed downlink radio resources for the MS.
Also if there is need to transfer delay sensitive real time traffic to both directions, downlink and uplink, the mobile station may request uplink radio resources only when the network assigns sending permission to the mobile station. This may cause a delay of variable amount of time, such as several seconds.
When allocating radio resources for downlink transfer, uplink radio resources cannot be allocated simultaneously because the uplink temporary block flow cannot be created without uplink packets. Thus it is possible, that the mobile station requests uplink radio resources but the network is unable to assign the requested radio resources.
Uplink radio resource deallocation is currently specified as follows: Every uplink RLC data block includes a countdown value (CV) field. It is specified in [1] that the CV shall be 15 when the mobile station has more than BS_CV_MAX (broadcast parameter) RLC data blocks left to be transmitted to the network. Otherwise the mobile station indicates to the network the number of remaining RLC data blocks with the CV field. The last RLC data block shall be sent to the network with the CV value set to ‘0’. Specification [1] defines also that once the mobile station has sent a CV value other than ‘15’, it shall not enqueue any new RLC data blocks meaning that the new RLC data blocks shall not be sent during the ongoing TBF. Once the network receives RLC data block with the CV field set to ‘0’, the TBF release procedures are initiated.
In uplink radio resource deallocation, the following problems may arise:
If delay sensitive real time data is transferred over radio interface according to current GPRS rules, the mobile station will have to establish several TBFs per session, because during the passive periods the mobile station has no RLC data blocks to send to the network and thus the CV value ‘0’ terminates the uplink TBF. Because the TBF setup procedure takes time, delay sensitive traffic cannot be transmitted with good quality. Also, there are no guarantees that free radio resources are always available when the mobile station requests uplink radio resources.
Downlink radio resource deallocation is currently specified as follows: Every downlink RLC data block includes a Final Block Indicator (FBI) field in the RLC header. The specification [1] defines that the network indicates to the mobile station the release of the downlink TBF by setting the FBI field to ‘1’. The network sets the FBI field to ‘1’ when it has no more RLC data blocks to send to the mobile station. After receiving RLC data block with FBI field set to ‘1’ the mobile station shall acknowledge to the network that it has received the FBI information. When the network receives the acknowledgement message, the TBF is released.
In downlink radio resource deallocation, the following problems may arise:
If delay sensitive real time traffic is transferred over radio interface according to current GPRS rules, the network would have to establish several TBFs per session, because during the passive periods the network has no RLC data blocks to send to the mobile station and thus the FBI value ‘1’ terminates the downlink TBF. Also, there are no guarantees that free radio resources are always available when the network tries to allocate downlink radio resources.
Problems also occur in assigning uplink and downlink sending permissions:
If delay sensitive real time data traffic is transferred on packet data channel/channels (PDCH), it is not guaranteed that adequate sending permissions are given in order to transfer the data, because the current network may not have unambiguous knowledge about delay sensitive data being transferred.
A further problem with the prior art specification is related to the feature that the network assigns transmission permissions for uplink and downlink directions independently, i.e. controls which mobile station receives data next and which mobile station may send data next. However, delay sensitive data, such as speech, has strict delay requirements. Consequently, whenever a delay sensitive data user has something to transmit, it must be able to do so in order to maintain an acceptable service level. If more than one users are allocated to the same packet data channel it is probable that at some point two or more users need to transmit simultaneously, and just one can be served on the channel. In speech conversations a large proportion of the connection time is silence. Thus it would be possible to statistically multiplex more than one speech user for one packet data channel. The prior art GPRS channel reservation system, however, is not elaborate enough to support this need. Therefore only one user of delay sensitive data transfer can be allocated for one packet data channel, which means that the use of the channel capacity is not optimised.
When the network notices that a mobile station wants to send delay sensitive data in the uplink direction the network reserves as much uplink resources to the mobile station as is requested. This naturally requires that the network has the required resources available. This may mean that the packet data channel is dedicated temporarily for a single mobile station in the uplink direction. During passive periods in uplink delay sensitive data transfer the network may assign uplink sending permissions of the allocated channels for other mobile stations. Since the mobile station transferring delay sensitive data reserves the uplink capacity of the packet data channel, other mobile stations that are allocated to the same packet data channel can not be assigned a sending permission to find out, whether they have data to send in the uplink direction. Also, if more than one mobile station allocated to the same packet data channel would need to send delay sensitive data at the same time, only one could be served. Therefore the network is forced to restrict the number of mobile stations transferring delay sensitive data according to the number of packet data channels in order to provide acceptable service quality.
As discussed above, in the present (E)GPRS (E=Enhanced) standards the MAC layer temporary block flows TBF are released after countdown procedure. This means that whenever the buffer is empty TBF is released. This means that the number of required TBF allocations, deallocations and reallocations during one user session can be high. This means that the effect of the problems discussed above may become severe.
Document [2] shows a solution for reducing the disadvantages of the prior art. This solution is based on a method to keep the TBF on also during idle periods when the transmission buffer is empty if the last RLC block has not been transmitted. The TBF is released only if a “release TBF” signaling message is received or if the idle (inactive) period lasts over a determined threshold period. This way many of the unnecessary TBF setups can be avoided and system efficiency can be improved.
However, with this solution the disadvantages can not be completely overcome. The “threshold period” can not be very long, because this would cause the use of the communication capacity to be inefficient, especially in situations of high communication loads. When TBF is on many network resources, such as temporary flow identifier (TFI), memory, etc., are reserved for the connection. Network needs also to allocate physical resources (eg. polling) for the MSs having TBFs for allowing them to transfer data. For this reason, TBFs should not be kept on unnecessarily long time periods.
But in case of applications which send small packets and wait for answers/acknowledgements to those packets, the TBFs will become released during the waiting period. For example all TCP based applications, such as Internet web browsing, begin the message exchange with three-way handshake where the following message is a response to the previous message. In such cases TBF is setup and released separately for each packet. Since TBF setup procedure loads control channels and is slow this procedure results in high signaling load, slow data transfer and low throughput.